基于EWT和ICA联合降噪的轴承故障诊断方法

针对轴承早期故障信号易被淹没于噪声中、故障特征难以提取的问题,提出一种经验小波变换(Empirical Wavelet Transform, EWT)与独立分量分析(Independent Component Analysis, ICA)的联合降噪方法。该方法依据峭度准则将经EWT分解得到的固有模态分量(Intrinsic Mode Function, IMF)重构后利用ICA进行盲源分离,有效抑制了振动信号中的噪声,使故障特征频率的能量幅值最大,从而识别故障特征。通过仿真分析和实际轴承早期故障的实验研究,表明该方法可明显削弱噪声干扰,突出故障频率成分。与EWT和包络谱结合的方法对比,信噪比提高了24.45%,能更清晰准确地提取故障特征信息,满足对轴承故障的诊断要求,为滚动轴承早期故障提取提供了一种方法。

Bearing Fault Diagnosis Method Based on EWT and ICA Combined Noise Reduction

Aiming at the problem that the early fault signal of bearing is easily submerged in noise and the fault feature is difficult to extract, an Empirical Wavelet Transform(EWT) and Independent Component Analysis(ICA) joint noise reduction method is proposed. The method reconstructs the Intrinsic Mode Function(IMF) obtained by EWT decomposition according to the Kurtosis criterion and then uses ICA to perform Blind Source Separation, which effectively suppresses the noise in the vibration signal and obtains the maximum energy amplitude of the fault characteristic frequency to identify fault characteristics. Through simulation analysis and experimental research on the actual bearing early failure, it shows that the method can significantly weaken the noise interference and highlight the fault frequency component. Compared with the method of combining EWT and envelope spectrum, the signal-to-noise ratio is improved by 24.45%, which can extract fault characteristic information more clearly and accurately, and meet the diagnostic requirements for bearing fault, which provides a method for early fault extraction of rolling bearing.